Bug: TypeError: dict.get(...).asNumber is not a function in n8n v2.3.5

[naejbravo]

When running the Tesseract node in n8n version 2.3.5, the execution fails with the following error:

NodeOperationError: dict.get(...).asNumber is not a function at ExecuteContext.execute (/home/node/.n8n/nodes/node_modules/n8n-nodes-tesseractjs/nodes/TesseractNode/TesseractNode.node.ts:425:12) at WorkflowExecute.executeNode (/usr/local/lib/node_modules/n8n/node_modules/.pnpm/n8n-core@file+packages+core_@opentelemetry+api@1.9.0_@opentelemetry+sdk-trace-base@1.30_ec37920eb95917b28efaa783206b20f3/node_modules/n8n-core/src/execution-engine/workflow-execute.ts:1045:8) at WorkflowExecute.runNode (/usr/local/lib/node_modules/n8n/node_modules/.pnpm/n8n-core@file+packages+core_@opentelemetry+api@1.9.0_@opentelemetry+sdk-trace-base@1.30_ec37920eb95917b28efaa783206b20f3/node_modules/n8n-core/src/execution-engine/workflow-execute.ts:1226:11) at /usr/local/lib/node_modules/n8n/node_modules/.pnpm/n8n-core@file+packages+core_@opentelemetry+api@1.9.0_@opentelemetry+sdk-trace-base@1.30_ec37920eb95917b28efaa783206b20f3/node_modules/n8n-core/src/execution-engine/workflow-execute.ts:1662:27 at /usr/local/lib/node_modules/n8n/node_modules/.pnpm/n8n-core@file+packages+core_@opentelemetry+api@1.9.0_@opentelemetry+sdk-trace-base@1.30_ec37920eb95917b28efaa783206b20f3/node_modules/n8n-core/src/execution-engine/workflow-execute.ts:2297:11

Environment:

• n8n version: 2.3.5

It seems like the node is using internal n8n methods (asNumber) that are no longer available or have changed in the latest n8n SDK versions.

[jreyesr]

Hello @naejbravo, and thanks for reporting this! Could you please provide the PDF that you're using, if it doesn't contain private information? Otherwise, try to create a small PDF and see if it also has the issue, perhaps one with a single image with a bit of text inside the image, and send the simplest PDF that also triggers the issue. That will be useful to see if the issue is general or if only certain PDFs trigger it.

The issue has to be somewhere here https://github.com/search?q=repo%3Ajreyesr%2Fn8n-nodes-tesseractjs%20asNumber&type=code, which is code that inspects each image that is embedded in the PDF. Perhaps the PDF was generated in a manner that isn't handled by the node.

[naejbravo]

Hi @jreyesr , thanks for your guidance.

I couldn’t share the original PDF because it contains private data, but I reproduced the setup with another file and confirmed the workflow works when the PDF is simple (3 pages, each page is basically an image with some text).

However, with the PDF I’m actually trying to process (a scanned document, 5 pages, and it seems to contain a lot more text/content), the workflow fails.

Questions:

Could the failure be related to the document complexity/size (scanned + lots of text/objects)?

Is there a way to enable debug logs or see more detailed execution logs in n8n to understand what is failing?

Any recommended settings or limits to check (memory/timeout/binary size)?

I’m new to n8n, so I might be missing something obvious. Any guidance on where to check logs or what to inspect first would be really helpful.

[jreyesr]

Hello! I don't think it's the amount of text in the images, because the failure happens before Tesseract is even invoked. It could certainly be the fact that it's scanned. Was your simpler document scanned or was it generated in Word or something similar? If it wasn't scanned, try to scan one with the same scanner+scanning program combo that you used for the failing document. Perhaps the scanning program stores images in an less-common way.

You can enable some debug logs by running N8N with the environment variable N8N_LOG_LEVEL=debug. This node does show additional logs when that level is set, so try that. Logs appear in the same terminal session if you run via CLI, or on the container's logs if you use Docker.